The installation industry runs on a combination of physical work and the work behind it that makes the physical work possible. Fitters on site, breakdown services, maintenance contracts, and behind that a layer of calculation, planning, work preparation and administration that determines whether the fitters' work runs smoothly or grinds to a halt. In many companies, a significant part of the non-executing hours goes to three things: making quotes based on drawings and specifications, aligning the planning of fitters and materials, and reworking based on what was found on site. That last part — the gap between what was quoted and what was actually needed — is in this sector often where margin leaks away and where clients become dissatisfied about additional work.
The outcome of a deal here is driven by circumstances that differ per company: the complexity of the type of installation, the degree to which work preparation is standardized, the dependence on the knowledge of a few experienced estimators, and how predictable the planning is in case of illness, breakdown or urgent work. Those are not abstract factors. They are the dimensions on which principals, consciously or unconsciously, choose between providers.
Calculation based on standard drawings and repetitive specifications is work that an AI application can take over in parts: recognizing patterns, counting quantities, pulling prices from historical data. For complex or unique installations, calculation remains at its core human work, with AI as an accelerator of the calculation work but not as a replacement for assessing risk. Planning is somewhere in between: schedules and material flows can be partly automatically aligned, but the exception — a breakdown, a sick leave notification, a supplier who does not deliver — requires a human to approve or intervene.
The effect of this shift is not in AI itself, but in what it does to the comparison between providers. Delivery time was long a matter of how many fitters a company could free up. Where planning largely runs itself, delivery time becomes less a result of staffing and more a result of how well the system processes unforeseen changes. Quote speed was a matter of how quickly an estimator freed up time; where calculation is largely automated, speed becomes a result of how clean the underlying data is, not of how hard people work. Companies that have already set this up this way compete on a different dimension than companies where calculation and planning still largely rest with individuals. That difference does not arise because one company is better or worse, but because the underlying tasks lend themselves differently to structuring — an existing ERP system, the degree of standardization in the type of work, and the willingness to set up oversight of what a system proposes all play a role in this.
That oversight is not a formality. A calculation that a system proposes must be approved or rejected by someone with a reason, especially in the case of deviations from the standard. Where that oversight is lacking, the risk shifts from slow quotes to incorrect quotes — a different problem, not necessarily a smaller one.
If calculation and planning become faster and more consistent for a portion of the providers, what a principal can compare on shifts. Response time to a request, the degree to which a quote matches the final invoice, and how predictable a schedule remains in the face of setbacks: these are dimensions that used to coincide with "how many people does this company have running around" and that are now becoming more loosely connected to staff size. A smaller company with well-organized work preparation can win on these points against a larger company where the same tasks still run manually.
That is not a statement about who is better. It is a shift in the dials on which companies compete, and that shift does not proceed evenly. Some companies have already largely organized calculation with systematic support and human approval on deviations; other companies still do everything based on the knowledge of a few people. For personnel decisions that might follow from such a shift, separate legal requirements apply; that is up to the employer, not up to a comparison of work processes.
This dynamic is not unique to installation companies. Similar shifts play out in the healthcare sector, where shift scheduling and handover are shifting under oversight, in the wholesale sector, where inventory management and order processing are the first to change, and in the manufacturing industry, where planning and quality control undergo the same shift. Also in the transport sector, where route planning was affected first, it can be seen that the dimension on which companies win shifts as soon as the underlying work changes.
The question of which work in a specific company can genuinely be taken over by AI — and which work remains human work — is answered by FTE TO AI's work scan per task, not in generalities. To determine against whom you are actually comparing, it helps to know what a peer group is and how you assemble one, and to see where no one in that peer group is scoring yet, an explanation of the white space analysis is relevant.
A good starting point is to name the dimensions on which you believe you win: response time, quote accuracy, planning stability, or something else that decides the deal in your practice. The free dimension check shows which of those claims can be defended with evidence and which still rest on assumption. The full competitive benchmark, with an evidence matrix per dimension, is under construction.